DDraceNetwork - developer

DDraceNetwork

Development / developer

Development discussion. Logged to https://ddnet.tw/irclogs/ Connected with DDNet's IRC channel, Matrix room and GitHub repositories — IRC: #ddnet on Quakenet | Matrix: #ddnet-developer:matrix.org GitHub: https://github.com/ddnet

Between 2023-06-13 00:00:00Z and 2023-06-14 00:00:00Z

Image attachment

04:00

I just want to build with MSVC

building works for me I think

04:03

unless it was a recent change that broke it

Ewan

I just want to build with MSVC

honestly. looks like it's just missing the data folder

Oh LOL

i always manually copy it

Yeah i know the issue exactly

04:15

it's the VS generator

04:15

i forgot it does that

04:16

vs puts the binary in a subfolder with build type

04:16

data is just put at the base of the build dir

ye

technically this is a cmakelists issue but i wouldn't blame anyone for it

i still always copy it and it's workable

are you not using cmake?

cmake does that with vs generator

^

i don't think we have other build system configured anyway.

Jupstar ✪

Pic or did not happen

dont believe me?

fokkonaut

dont believe me?

Yes

I can show u later but why

Why would I believe someone

believe what

06:29

Ah screenburn ye I also doubt that haha

fokkonaut

I can show u later but why

Nonono, show everyone

06:32

I want to see, too

How cool would it be if maps had area boxes that could color correct the screen? This is post processing BTW Pulsar called it useless but ngl I think this would make maps feel more immersive

2

For a thing like that to look consistent, we'd need to introduce some concept of lighting

only downside is that people dont know what color theory and proper lighting is so there would be a lot of

Image attachment

Jupstar ✪

For a thing like that to look consistent, we'd need to introduce some concept of lighting

maybe. i mean, i only see it as boxes that work similar to sounds

06:54

idk how lighting in ddnet would work but id love to know (edited)

<Jupstar> mhh yeah it's not that easy. I thought about it when i thought about a good looking 2D ray tracer (which obviously makes no sense in 2D)

06:57

<Jupstar> ambient light is ez. is the question if it should affect certain areas only

06:58

<Jupstar> for caves to look dimmer we'd need some concept like a sun. but that makes 0 sense in 2D

06:59

<Jupstar> we could analyze how much of the sun is blocked while rendering before rendering the tee

06:59

<Jupstar> so a cave would require to have a background tile layer

07:00

<Jupstar> towards cave entries more sun would generally be in the scene

when i think of lighting i think about glow tiles too basically like, a tileset with a set color and emittion amount saying this because glow mapres suck so hard there's always creases and it never looks natural

Image attachment

<Jupstar> yeah but obv that also doesnt really make sense.. e.g. are there tiles that can completely block the emitting light rays? or does it always leak into the tiles

oh yeah good point. i was going to say it would be layer determined but i guess if the emittion is big enough there should be ways to block the lighting

Image attachment

1

lel, didn't know the last one

That last one is a cute idea, I wonder if it's sane

I heard it works

10:14

not sure if practical though

you can also pick random points in a square that are also in the quarter circle to get pi

10:20

Anyone know how to test C code similarly to googletest that is c++ only?

you can use C++ to test C code

isn't there a risk to change behavior?

10:22

if i compile my code with g++ instead of gcc, i got loads of errors

no, I mean just compiling the test code with g++

Note that i will need to compile C code from students, so i think a pure C solution is preferable

does it have a given interface?

ah, so i compile my test with g++ and retrieve only their code through a library maybe?

doesn't need to be a library

they should have 2 functions:

int aes_encrypt(char* data, int size, char* key, int keysize);
int aes_decrypt(char* data, int size, char* key, int keysize);
// OU avec AES CBC implémenté
int aes_encrypt(char* data, int size, char* key, int keysize, int mode);
int aes_decrypt(char* data, int size, char* key, int keysize, int mode);

10:24

ill try that then

yes, declare that as extern "C" in your header file and you should be good

10:24

compile the student's code with -c to an object file

10:25

and then link it with your test cde

I was thinking of them sending me their header and also 2 static libraries: one normal, and one with the sanitizer, then i can just link with my testcode

hmm. you don't want to have the source code?

Virus

but yea, that'd work, too

nah, they will all have different makefiles and their own test files

10:26

so i wouldn't know which .o to get

10:27

it will also be a live evaluation, so i can't take 5 minutes for each group to find correct files and stuff (edited)

Make files for single algorithm? Xd

ah ^^

Or do they use extern stuff

I will ask them for extern C header and static lib, seems easier so that i can just link directly

<Jupstar> the static lib can also have deps

they won't have

<Jupstar> then u can also take their source code

https://discord.com/channels/252358080522747904/293493549758939136/1118124259970138253

10:30

we didn't standardize anything except global API, so i don't wanna fumble through their code

10:30

live evaluation

<Jupstar> then u need to take shared libraries

why shared? static works fine too

this would be way easier with rust, cargo, normalized way of doing things already and builtin unit testing

10:31

justatest

<Jupstar> then i hope for you that they dont have external deps

Ryozuki

this would be way easier with rust, cargo, normalized way of doing things already and builtin unit testing

i don't make the teaching program

10:32

there will be none, it's just basic c stuff with file IO and AES algorithm

why not just provide urself a project skeleton

10:32

with a makefile u made

10:32

good old make

<Jupstar> and then u could take their source code again^^

because we also evaluate this

what is "this"

their own makefile

ah

10:33

so in computer design class u evaluate makefile skills?

10:33

xd

it's not computer design class, it's C project class

<Jupstar> its plain c

actually, it's math-info project

C project class lol

<Jupstar> u dont really need makefiles lmao

so they already made AES with Haskell, now with C

nice

10:33

should be easier

10:33

xd

10:33

idk haskell

10:34

do processors have aes intrinsics

10:34

like instructions

yes

do they use em in c

10:34

can u do that with haskell

there's a possible extension that they can use, i proposed it to some advanced groups

oh

im only supervising the c part

i also think the linux kernel provides a interface for that

10:35

if u enable it when compiling

10:35

https://www.kernel.org/doc/html/v4.15/crypto/architecture.html

10:35

but i understand u want them to impl it itself

10:35

xd

<Jupstar> rust would link extern "Rust"

10:35

<Jupstar> !rust *

@Ryozuki

extern "Rust" exists i think

<Jupstar> yes

https://www.intel.com/content/www/us/en/docs/intrinsics-guide/index.html#othertechs=AES

Intel® Intrinsics Guide

Intel® Intrinsics Guide includes C-style functions that provide access to other instructions without writing assembly code.

10:36

AES extension on x64

there also exists extern "x86-something"

introduced in 2009 apparently or 2008

<Jupstar> no wonder x86 is so huge

we also ask them to compare performances, own implementation are usually at 1MB/s at start, then they manage to reach 10MB/s. With AES extension, they reach 1GB/s

<Jupstar> i'd aim for 2GB/s

10:37

<Jupstar> else im unhappy

not possible

<Jupstar> that's what u say

type HandlerFunc = extern "x86-interrupt" fn(_: InterruptStackFrame);

maybe possible if you cipher independent block

10:37

so that you break dependencies

<Jupstar> @Chairn first i'd buy you a new PC. then add multi threading

10:38

<Jupstar> and then it's 16GB/s

xd troll

10:38

@Chairn imagine having jupstar as student

10:39

nagging all day

10:39

i problably would nag all day

10:39

smth C this smth C that

10:39

ub here ub there

jupstar buying pcs for people is such a jupstar move

<Jupstar> i am defs a horrible student

10:40

<Jupstar> the funny thing is. i in fact did smth like that

justatest

1

see? jupstar moves

i also nagged to the teacher

<Jupstar> i had to create a java program. and just for fun aimed for most effective perf :D

xd

<Jupstar> i think he didnt understand the code xD

10:40

<Jupstar> man i should search my old tasks

10:41

<Jupstar> i bet i did weird stuff, bcs i always hated the homeworks

10:41

<Jupstar> to death

10:41

<Jupstar> boring af

true

10:41

homework is shit to do

<flamer> aaa

opsi

<Jupstar > OMG

12:03

<Jupstar > chillerdragon

12:03

<Jupstar > u didnt

12:03

<Jupstar > u didnt even tell me

12:03

<Jupstar> but its still a bit buggy, but nice progress "Jupstar is typing ..." after reload

yes i just fixed the stuck bug

12:22

lemme deploy

<ChillerDragon> its now fixed

12:23

<ChillerDragon> the problem was the "is typing" state is tracked by the backend on a trust me bro basis from the client

12:24

<ChillerDragon> the client sends when it stops typing :D but that doesnt happen if you timeout without saying u stopped typing now its kinda fixed

12:24

<ChillerDragon> you gotta relogin since i restarted the backend

[ddnet-web:master] 1 new commit

2d85b36 Add back Turkey - def-

5

<ChillerDragon> !jap https://zillyhuhn.com/cs/.1686669397.png

Embedded image

@Jupstar ✪

<ChillerDragon> pingers from other channel poggers2

poggers2

15:17

<ChillerDragon> @heinrich5991 when bridge #off-topic?

Chillerdragon is creating a next-gen webclient as discord alternative?

ye its a web frontend for irc that looks like discord

15:20

but ffs i broke something really bad with this new feature xd

<ChillerDragon> !pig

pog

<ChillerDragon> ok nice fixed

[ddnet-web:master] 1 new commit

881bba3 Frontpage video - def-

https://llvm.org/docs/CodingStandards.html#anonymous-namespaces (edited)

16:17

possible improv?

16:17

ops wrong link before

[ddnet/ddnet-discordbot] Pull request opened: #48 Add support for multiple categories

I hope this works, I'll merge when/if it's needed

Why is there no abuse email for googles ip range? Do they get special treatment because monopoly?

16:32

My ip is used in one spoofed attack and hetzner is up my ass because they got abuse mail. We get a gbps-ful of traffic from 8.8.4.4 I want to complain too

It's spoofed traffic anyway, so why complain?

16:37

Popular to spoof since people whitelist those ips in their firewall for DNS

16:38

abuse emails seem to be a joke anyway, I mostly get back "address doesn't exist" or "mailbox is full"

idk we are being annoyed, I want to share our pain with them

!c nice xd

ChillerDragon

You sure 8.8.4.4 is not just responses to your dns requests? Don’t think a response counts as attack

Sure, we're just sending multi-GB/s of DNS requests and google is failing to limit us

Ok fair

17:22

Wait hetzner is sending you emails because your ip was used in a spoofing attack? A attack that was fully out of your control? That doesn’t make much sense. Or do you mean these emails about you running services that might be used in reflection attacks? Because those we get at work and there is something we can do about it so it’s making sense imo.

17:25

So it’s probably someone ddosing you and spoofing google ip to not get ip blacklisted? (@deen)

Sure, we're just sending multi-GB/s of DNS requests and google is failing to limit us

17:26

Imagine using evil corp dns anyways… real OGs use duckDNS

[ddnet/ddnet] Pull request opened: #6736 Fix editor save dialog entries not selectable with mouse anymore

Regression from #6530.

Checklist

[X] Tested the change ingame
[ ] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[ ] Written a unit test (especially base/) or added coverage to integration test
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect existing maps
[ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddn...

@Learath2 just wanted to comment, we are back to cxx with the bindings

18:51

some stuff turned weird xD

bruh

the bad thing with cxx is that u make unique pointers everywhere, so allocations everywhere for everything

Ryozuki

the bad thing with cxx is that u make unique pointers everywhere, so allocations everywhere for everything

You technically don't have to, ur also allowed raw pointers across the boundary, and you can define non-opaque structs that can go across the boundary

my coworker mentiones it might be a red zone thing

18:53

but we are not sure

18:53

a pointer address changed doing stuff, with the context allocated from rust

18:53

we couldnt figure out why

wym a pointer address changed?

Learath2

You technically don't have to, ur also allowed raw pointers across the boundary, and you can define non-opaque structs that can go across the boundary

yeah but in mlir u can get Value(s) and Type(s) and Location(s) by value a lot

18:54

since they are trivially copiable

18:54

or smth

18:54

llvm returns them by val

Learath2

wym a pointer address changed?

the Context in mlir is just a unique ptr inside

18:54

when we created a Location with it

18:54

and Location has a getContext method too

18:55

the getcontext returned a context with a pointer with a 100 byte offser more or less

18:55

it changed +-50

18:55

offset*

18:55

anyway it was bound to be a disaster anyway xd

18:55

allocating in rust and using in c++

18:56

there must be smth weird

Learath2

You technically don't have to, ur also allowed raw pointers across the boundary, and you can define non-opaque structs that can go across the boundary

u can use raw pointers if the api returns pointers/refs ye

18:57

but the api mostly just returns by value

18:57

in mlir there are operations, they almost always have results, these results return Value by value

18:58

and thus thats a allocation for every operation result

18:58

a operation can also have multiple results

18:58

anyway

i should make more ddnet feature concept images (edited)

Voxel

i should make more ddnet feature concept images (edited)

add them to the wiki if u want too

Voxel

i should make more ddnet feature concept images (edited)

u should make a beter ddnet logo

i allow the wiki to be ur ddnet concept art storage

19:00

xd

Jupstar ✪

u should make a beter ddnet logo

its impossible

19:00

the current logo is perfect

1

^^

Ryozuki

add them to the wiki if u want too

really? what page?

no page

19:02

u can simply upload

19:02

https://wiki.ddnet.org/wiki/Special:Upload

19:02

@Voxel u can show em in ur user page if u want

19:02

it also appears here https://wiki.ddnet.org/wiki/Special:ListFiles

Special:ListFiles

19:03

@Voxel oh

19:03

Image attachment

19:03

i uploaded some of urs in the past

19:03

xd

19:04

https://wiki.ddnet.org/wiki/File:Voxel_does_a_pull_request_part2.png

File:Voxel does a pull request part2.png

19:04

lmao

oh LOL

Jupstar ✪

u should make a beter ddnet logo

what about the current logo is so bad? (edited)

[ddnet/ddnet] Pull request opened: #6737 Refactor `str_copy` usages and buffer sizes in editor

Use templated str_copy function in most cases to specify the correct buffer size. Fixes two cases where the buffer size was hard-coded to a number. Fix buffer size not being checked when copying auto mapper configuration name. Loading an auto mapper configuration with a name longer than 128 bytes could cause an out-of-bounds write. Increase sizes of some text buffers that were potentially too small.

Checklist

[X] Tested the change ingame
[ ] Provided screenshots if it...

Voxel

what about the current logo is so bad? (edited)

do u also like it? i think it looks cheap i just feel like it doesnt represent teeworlds. Choice of colors, the default tees I like the blue rings tho xd

Jupstar ✪

do u also like it? i think it looks cheap i just feel like it doesnt represent teeworlds. Choice of colors, the default tees I like the blue rings tho xd

i cant think of anything better

19:14

i mean the only thing i WANT to improve are the menu images

19:14

Image attachment

1

progress so far

Image attachment

no.

19:17

cute twinbop

can it twerk?

Anime.pdf

can it twerk?

Image attachment

Voxel

progress so far

anyways im gonna continue this

Twinbop

Voxel

Click to see attachment 🖼️

deen

operating system. Yes, sounds like the gfx troubleshooting would help you

@deen i already installed the drivers and everything was working fine again. Until now that i tried to save a map in editor, then it suddenly crashed my game and now it says that i need to update my GPU drivers. How do I do that?

RealTaco

@deen i already installed the drivers and everything was working fine again. Until now that i tried to save a map in editor, then it suddenly crashed my game and now it says that i need to update my GPU drivers. How do I do that?

You can find the new drivers on the amd or nvidia website

20:23

or use older opengl version in settings

[ddnet:master] 2 new commits

b13e807 Fix editor save dialog entries not selectable with mouse anymore - Robyt3 fa57258 Merge #6736 - bors[bot]

20:50

[ddnet/ddnet] Pull request closed: #6736 Fix editor save dialog entries not selectable with mouse anymore

@Jupstar ✪ https://www.anandtech.com/show/18915/amd-expands-mi300-family-with-mi300x-gpu-only-192gb-memory

AMD Expands AI/HPC Product Lineup With Flagship GPU-only Instinct M...

21:04

lets buy this as ddnet

21:04

for research purposes

21:04

Image attachment

generate ddnet maps with it

itd only be good at gores

[ddnet:master] 2 new commits

bbe9f23 Refactor str_copy usages and buffer sizes in editor - Robyt3 1830867 Merge #6737 - bors[bot]

21:30

[ddnet/ddnet] Pull request closed: #6737 Refactor `str_copy` usages and buffer sizes in editor

[ddnet/ddnet] Issue opened: #6738 src/engine/server/databases/mysql.cpp(349): index out of bounds

Happend on RUS server Dump: ``` 2023-06-13 16:21:04 I sql: SQLite statement: INSERT OR IGNORE INTO record_race_backup( Map, Name, Timestamp, Time, Server, cp1, cp2, cp3, cp4, cp5, cp6, cp7, cp8, cp9, cp10, cp11, cp12, cp13, cp14, cp15, cp16, cp17, cp18, cp19, cp20, cp21, cp22, cp23, cp24, cp25, GameID, DDNet7) VALUES ('HDP_OnOff', 'Dima4qlc', DATETIME('2023-06-13 16:21:04', 'utc'), 1169.00, 'RUS', 15.82, 28.46, 51.04, 72.74, 89.24, 149.38, 162.70, 0.00, 0.00, 0.00, 0.00, 0.00, 0.00,...

[ddnet/ddnet] Pull request opened: #6739 Add templated `str_append` function for arrays with fixed size, add tests, ensure proper buffer size is used with DDNet server filter

Checklist

[X] Tested the change ingame
[ ] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[X] Written a unit test (especially base/) or added coverage to integration test
[X] Considered possible null pointers and out of bounds array indexing
[X] Changed no physics that affect existing maps
[ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresss...

deen

or use older opengl version in settings

Now what

Image attachment

Remove the custom skins you have

^ or archive them in a seperate folder

22:30

just not in the folder ddnet reads from

[ddnet/ddnet] Pull request opened: #6740 Don't remove the `.map` suffix in editor file dialog

This otherwise breaks file loading because the file extension is expected in this input buffer even when the input is not rendered. Regression from #6736.

Checklist

[X] Tested the change ingame
[ ] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[ ] Written a unit test (especially base/) or added coverage to integration test
[ ] Considered possible null pointers and out of bounds array indexing
[ ...

RealTaco

Now what

Remove some Skins or Mapres, whatever part is bigger. I recommend getting rid of 4k Assets as priority

22:51

Also main reason why I don't wanna include a "download all" button on skins.tw

since when was it EEnumName instead of ENUM_NAME? i feel like i read on the wiki it was the latter

[ddnet:master] 5 new commits

9a57def Add templated str_append function for arrays with fixed size - Robyt3 410ccda Move templated str_copy next to basic function - Robyt3 3069733 Add tests for str_append and for templated str_copy - Robyt3 3ad581a Ensure proper buffer size is used with DDNet server filter - Robyt3 0d45120 Merge #6739 - bors[bot]

23:44

[ddnet/ddnet] Pull request closed: #6739 Add templated `str_append` function for arrays with fixed size, add tests, ensure proper buffer size is used with DDNet server filter

Exported 293 message(s)